Anomaly classification: Across the whole Dutch highway system there is a large variety of anomalies in the data. In order to monitor the condition of the road it is important that we accurately classify these anomalies, separating damage from other outliers. However, it is challenging to accurately classify different structures that have similar visual features. For example, separating cracks in the asphalt from local repairs, pavement joints, road markings, or long stretches of raveling. In this project, you will investigate (deep learning) solutions to perform automatic detection and classification of such anomalies. Pavement type classification using deep learning: Not all asphalt is the same, there are actually several different types of asphalt used to pave the Dutch highways. It is difficult to accurately estimate the amount of general wear accurately without knowing the exact asphalt type. A rough porous type of asphalt might look damaged if you would think it was a fine, dense piece of asphalt. During this project, you will create an algorithm to classify different types of asphalt, or even estimate the approximate stone-size distribution within the asphalt.